Solving Kepler's equation — Newton-Raphson iteration: Visual walkthrough

We are chasing this one boxed sentence:

$M = E - e sin E$

But before that box means anything, we must draw the shapes, name the angles, and count the areas. Let's go from zero.

Step 1 — Draw the orbit and its shadow-circle

WHAT. We draw the elliptical orbit (the squashed circle the planet actually travels) and then draw a perfect circle around it that just touches its widest points. That circle is the auxiliary circle.

WHY. An ellipse is awkward to measure angles on — its "radius" changes everywhere. A circle is friendly: every point sits the same distance from the centre. The trick of the whole derivation is: do the geometry on the easy circle, then squash the answer down onto the ellipse. (This same auxiliary circle is the home of the eccentric anomaly.)

PICTURE. Look at the figure. The mint ellipse has:

semi-major axis $a$ — half the long width (centre to the far edge, along the wide direction),
semi-major axis label and semi-minor axis $b$ — half the short height.

The lavender circle has radius exactly $a$ . The centre is the dot marked $C$ . Off to one side sits the focus $F$ — the special point the Sun lives at, not the centre.

Step 2 — Name the steady clock-angle $M$

WHAT. We invent a pretend angle that grows at a perfectly constant rate as time ticks. Call it $M$ , the mean anomaly.

WHY. The real planet does not move at constant angular speed (Kepler's 2nd law: it hurries near the Sun, dawdles far away). That makes "where is it at time $t$ ?" hard. So we first build something that is uniform in time — a mental clock-hand — and later connect it to the real position. $M$ is defined so that it sweeps a full $2 π$ in exactly one orbital period $T$ (see Mean Motion and Orbital Period).

PICTURE. The butter-coloured clock-hand sweeps at a steady rate. In equal time-slices it opens equal angles — the little arcs $Δ M$ are all the same size. That uniformity is the entire reason $M$ exists.

Step 3 — Name the geometric angle $E$ on the circle

WHAT. Take the planet on the ellipse. Slide straight up (parallel to the short axis) until you hit the auxiliary circle. The angle at the centre $C$ to that circle-point is the eccentric anomaly $E$ .

WHY. We need a handle on the planet's real position, but measured on the friendly circle. $E$ is that handle. Crucially $E$ is measured from the centre, while the true angle of the planet is measured from the focus — that mismatch is where the whole equation's "correction term" will come from.

PICTURE. Notice the vertical dashed line: the planet $P$ on the ellipse and the point $Q$ on the circle share the same horizontal position. The angle $∠ QC X$ (from the wide axis, at centre $C$ ) is $E$ . The height of $Q$ is $a sin E$ ; its horizontal offset from centre is $a cos E$ .

Step 4 — Equal areas means equal time

WHAT. By Kepler's 2nd law, the line from the focus to the planet sweeps equal areas in equal times (see Kepler's Laws of Planetary Motion). So the area swept since perihelion is a fixed fraction of the total ellipse area — and that fraction is exactly $M /2 π$ from Step 2.

WHY. This is what converts a time statement into an area statement. Time is invisible; area we can compute with geometry. This is the pivot of the whole derivation.

PICTURE. The shaded coral wedge is the area the focus-line has swept from perihelion out to the planet. Its size, divided by the whole ellipse, equals the fraction of the period elapsed.

Step 5 — Compute the swept area on the circle: pie slice minus triangle

WHAT. Do the swept area on the circle instead. The region swept from perihelion, measured from the focus, is a pie slice about the centre minus a triangle that hangs off toward the focus.

WHY. Circle areas are trivial: a pie slice of angle $E$ in a circle of radius $a$ has area $\frac{1}{2} a^{2} E$ . But our wedge is measured from the focus, not the centre — the focus is shoved over by $c = a e$ . That offset carves off a triangle. Subtract it.

PICTURE. The lavender pie slice $\frac{1}{2} a^{2} E$ is the full wedge from the centre. The coral triangle $C$ – $F$ – $Q$ must be removed because our real reference point is $F$ , sitting $a e$ to the side. That triangle has base $a e$ (centre-to-focus) and height $a sin E$ (height of $Q$ ).

Step 6 — Squash the circle area onto the ellipse

WHAT. Every point on the circle drops to the ellipse by shrinking its height by the factor $b / a$ . Shrinking every height by $b / a$ shrinks every area by $b / a$ too.

WHY. Areas scale by the same factor as the one direction you squash (widths are untouched, heights all multiply by $b / a$ , so the product — area — multiplies by $b / a$ ). This lets us borrow the easy circle-area result and land it on the hard ellipse.

PICTURE. The tall lavender circle-wedge on the left, and the same wedge squashed flat into the coral ellipse-wedge on the right. Every vertical strip lost the same fraction $b / a$ of its height.

Step 7 — Assemble and watch the letters cancel

WHAT. Put the squashed area into the area-law of Step 4 and simplify.

WHY. This is the final algebra — and it is beautiful, because every $a$ and $b$ cancels, leaving a pure relation between angles.

PICTURE. The cancellation, drawn: $\frac{1}{2} ab$ on top and $π ab$ on the bottom both carry the same $ab$ ; strike them out and only $(E - e sin E)$ and $2 π$ survive.

$\frac{\frac{1}{2} ab ( E - e sin E ) swept ellipse area}{whole ellipse π ab} = \frac{M}{2 π} ⟹ \frac{E - e s i n E}{2 π} = \frac{M}{2 π} ⟹ M = E - e sin E$

Step 8 — Edge cases: the equation must survive the extremes

WHAT. Check the corners: a circle, the two apsides, and the near-parabolic limit.

WHY. A law you trust must not break at $e = 0$ or $e \to 1$ . Let's stress-test it.

PICTURE. Three mini-panels: a perfect circle (left), the perihelion/aphelion line where the correction dies (middle), and the stretched near-parabola where the correction rages (right).

The one-picture summary

Everything above, compressed: the circle gives the pie-slice, the focus carves the triangle, the squash by $b / a$ drops it onto the ellipse, and the area-law equates it to the clock fraction $M /2 π$ — leaving $M = E - e sin E$ .

Recall Feynman retelling — say the whole walkthrough in plain words

We wanted to connect the time on a planet's journey to where it actually is. Time is smooth and boring, so we invented a pretend clock-hand called $M$ that sweeps at perfectly even speed — after a fraction of the orbit's period, it has swept that same fraction of a full turn.

But the planet cheats: near the Sun it races, far away it crawls, sweeping equal areas in equal times. So instead of chasing its messy angle directly, we drew a big helper circle around the squashed orbit and marked the angle $E$ from the centre up to the planet's shadow on that circle.

To measure the area the planet swept, we cut a clean pie slice out of the circle (easy: $\frac{1}{2} a^{2} E$ ), then sliced off a triangle because the Sun sits off-centre by $a e$ — that triangle is $\frac{1}{2} a^{2} e sin E$ . Slice minus triangle gives $\frac{1}{2} a^{2} (E - e sin E)$ . We then squashed the whole circle-area down onto the ellipse by the factor $b / a$ , so the real swept area is $\frac{1}{2} ab (E - e sin E)$ .

Finally we said: (area swept)/(whole ellipse $π ab$ ) must equal (clock fraction) $M /2 π$ . All the $a$ 's and $b$ 's cancel, and out pops the jewel: $M = E - e sin E$ . The $- e sin E$ is just the memory of that off-centre triangle. And because you can't peel $E$ back out of that equation with algebra, we have to guess-and-correct — which is where Newton-Raphson takes over.

Recall Quick self-test

Why does the correction term vanish at perihelion and aphelion? ::: Because there $sin E = 0$ (at $E = 0$ and $E = π$ ), so the off-centre triangle has zero height and area — nothing to subtract. What geometric object is the $- e sin E$ term? ::: The triangle from centre $C$ to focus $F$ to the circle-point $Q$ , with base $a e$ and height $a sin E$ . Why work on the circle instead of the ellipse directly? ::: Circle sectors have the trivial area $\frac{1}{2} a^{2} E$ ; we compute there and squash by $b / a$ onto the ellipse.

Parent: Solving Kepler's equation — Newton-Raphson iteration · Related: Eccentric Anomaly and the Auxiliary Circle · True Anomaly and the Orbit Equation · Mean Motion and Orbital Period